Multi-Strategy Adversarial Continual Graph-Feature Learning for Robust Financial Fraud Detection

Authors

  • Victor Choi Department of Computer Science, North Carolina State University, USA
  • Nina Sullivan College of Computing & Informatics, Drexel University, USA
  • Olivia Tan Department of Computer Science, Florida State University, USA

DOI:

https://doi.org/10.66421/int.j.datasci.intell.syst75

Keywords:

financial fraud detection, graph-inspired features, continual learning, adversarial perturbations, strategy selection, synthetic signatures

Abstract

Robust financial fraud detection requires online adaptation to changing transaction patterns and deliberate manipulation of relational indicators. Building on the strategic-drift and topology-aware memory principles of GT-ACGL, we develop a computationally lightweight continual-learning approach that combines ego-network-style relational signatures, class-balanced structural replay and prioritized attack-view augmentation for an online classifier. Seven sequential training policies are evaluated over six chronological windows under structural dilution, camouflage, burst suppression and compound perturbations. Across five random seeds, prioritizing dilution with replay raises compound-attack F1 to 0.925, compared with 0.506 for fine-tuning and 0.846 for uniform multi-view replay, and an eight-seed sweep shows that the advantage persists across six attack intensities (up to 12.8 percentage points; paired p = 5.76×10⁻⁵ at the reference intensity). These results identify attack-view allocation as a consequential design choice for inexpensive continual fraud screening and provide a reproducible basis for event-level extensions.

References

Rossi, E., et al., “Temporal Graph Networks for Deep Learning on Dynamic Graphs,” arXiv:2006.10637, 2020. https://arxiv.org/abs/2006.10637

Xu, D., et al., “Inductive Representation Learning on Temporal Graphs,” in ICLR, 2020. https://arxiv.org/abs/2002.07962

Weber, M., et al., “Anti-Money Laundering in Bitcoin: Experimenting with Graph Convolutional Networks for Financial Forensics,” arXiv:1908.02591, 2019. https://arxiv.org/abs/1908.02591

Zhou, F. and Cao, C., “Overcoming Catastrophic Forgetting in Graph Neural Networks with Experience Replay,” in AAAI, vol. 35, pp. 4714–4722, 2021. https://doi.org/10.1609/aaai.v35i5.16602

F. Zhou, “Moment-constrained side-force allocation for wind-assisted merchant ships: Hydrodynamic performance of retractable anti-leeway fins and flap rudders,” Innovation and Technology Studies, vol. 3, no. 1, pp. 27–41, 2026.

Lee, D., Lee, J., and Shin, K., “Spear and Shield: Adversarial Attacks and Defense Methods for Model-Based Link Prediction on Continuous-Time Dynamic Graphs,” in AAAI, vol. 38, pp. 13374–13382, 2024. https://doi.org/10.1609/aaai.v38i12.29239

J. Wang, Y. Ge, J. Liu, and B. Jiang, “Early warning of liquidity stress in U.S. banks with seasonally aligned frailty hazards and delayed-feedback conformal alarms,” Journal of Trends in Finance and Economics, vol. 1, no. 1, pp. 85–100, 2024.

Perdomo, J., Zrnic, T., Mendler-Dünner, C., and Hardt, M., “Performative Prediction,” in ICML, PMLR 119, pp. 7599–7609, 2020. https://proceedings.mlr.press/v119/perdomo20a.html

Cong, W., et al., “Do We Really Need Complicated Model Architectures for Temporal Networks?,” in ICLR, 2023. https://arxiv.org/abs/2302.11636

Zügner, D. and Günnemann, S., “Adversarial Attacks on Graph Neural Networks via Meta Learning,” in ICLR, 2019. https://arxiv.org/abs/1902.08412

Zügner, D., Akbarnejad, A., and Günnemann, S., “Adversarial Attacks on Neural Networks for Graph Data,” in IJCAI, pp. 6246–6250, 2019. https://doi.org/10.24963/ijcai.2019/872

Kumar, S., Zhang, X., and Leskovec, J., “Predicting Dynamic Embedding Trajectory in Temporal Interaction Networks,” in KDD, pp. 1269–1278, 2019. https://arxiv.org/abs/1908.01207

Mujkanovic, F., Geisler, S., Günnemann, S., and Bojchevski, A., “Are Defenses for Graph Neural Networks Robust?,” in NeurIPS, vol. 35, pp. 8954–8968, 2022. https://proceedings.neurips.cc/paper_files/paper/2022/hash/3ac904a31f9141444009777abef2ed8e-Abstract-Conference.html

Kazemi, S. M., et al., “Representation Learning for Dynamic Graphs: A Survey,” J. Mach. Learn. Res., vol. 21, pp. 1–73, 2020. https://jmlr.org/papers/v21/19-447.html

Cheng, D., Zou, Y., Xiang, S., and Jiang, C., “Graph Neural Networks for Financial Fraud Detection: A Review,” Front. Comput. Sci., vol. 19, 199609, 2025. https://doi.org/10.1007/s11704-024-40474-y

Liu, Y., Ao, X., Qin, Z., et al., “Pick and Choose: A GNN-Based Imbalanced Learning Approach for Fraud Detection,” in WWW, pp. 3168–3177, 2021. https://doi.org/10.1145/3442381.3449989

Wang, J., Zhu, W., Song, G., and Wang, L., “Streaming Graph Neural Networks with Generative Replay,” in KDD, pp. 1878–1888, 2022. https://doi.org/10.1145/3534678.3539336

Zhang, X., Song, D., and Tao, D., “Sparsified Subgraph Memory for Continual Graph Representation Learning,” in ICDM, pp. 1335–1340, 2022. https://doi.org/10.1109/ICDM54844.2022.00177

J. Chen, X. Zhang, and X. Han, “Explainable identification of tax-related financial reporting anomalies in U.S. public companies: Cross-fitted tax-consistency residuals and Owen-value Shapley propagation,” World Journal of Economics and Business Research, vol. 1, no. 1, pp. 33–45, 2023.

Li, H., Zhang, Z., Wang, X., and Zhu, W., “Learning Invariant Graph Representations for Out-of-Distribution Generalization,” in NeurIPS, vol. 35, pp. 11828–11841, 2022. https://papers.nips.cc/paper_files/paper/2022/hash/4d4e0ab9d8ff180bf5b95c258842d16e-Abstract-Conference.html

Souza, A., Mesquita, D., Kaski, S., and Garg, V., “Provably Expressive Temporal Graph Networks,” in NeurIPS, vol. 35, pp. 32257–32269, 2022. https://arxiv.org/abs/2209.15059

D. Teng, “RSRBench: Reproducibility and performance benchmarking for regional seismic risk workflows,” Academic Journal of Architecture and Civil Engineering, vol. 3, no. 1, pp. 40–52, 2025.

Liu, R., Gao, J., Zhang, J., Meng, D., and Lin, Z., “Investigating Bi-Level Optimization for Learning and Vision from a Unified Perspective: A Survey and Beyond,” IEEE TPAMI, vol. 44, pp. 10045–10067, 2022. https://arxiv.org/abs/2101.11517

Zeng, H., Zhou, H., Srivastava, A., Kannan, R., and Prasanna, V., “GraphSAINT: Graph Sampling Based Inductive Learning Method,” in ICLR, 2020. https://arxiv.org/abs/1907.04931

Kirkpatrick, J., et al., “Overcoming Catastrophic Forgetting in Neural Networks,” Proc. Natl. Acad. Sci., vol. 114, pp. 3521–3526, 2017. https://doi.org/10.1073/pnas.1611835114

X. Yue, J. Yang, R. Zhu, and Q. Xu, “Adaptive Temporal Graph Neural Networks for Real-Time Financial Fraud Detection under Concept Drift,” Highlights in Business, Economics and Management, vol. 69, pp. 239–250, 2026.

J. Ding, Z. Shen, and W. Liu, “Game-Theoretic Cost-Sensitive Adversarial Training for Robust Cloud Intrusion Detection Against GAN-Based Evasion Attacks,” Applied Sciences, vol. 16, no. 8, Art. no. 3944, 2026, doi: 10.3390/app16083944.

Downloads

Published

2026-10-19